Training mixture models and DP mixture models on well-separated 2D Gaussian blobs.
Initialization for Mixtures of Gaussians
Variational coordinate descent for Mixture of Gaussians
Variational with birth and merge proposals for DP mixtures of Gaussians
Warm Starting from a Previously Trained Mixture of Gaussians
Variational training of DP mixture models on thousands of points from a single 1D Gaussian with mean 0 and variance 1.
Variational with merge and delete proposals for DP mixtures of Gaussians
Variational training of DP mixture models on classic dataset on eruption times from Old Faithful geyser in Yellowstone National Park.
Variational training for Mixtures of Gaussians
Variational with merge and delete proposals for DP mixtures of Gaussians
Demonstration of Sparse Responsibilities for Mixtures of Gaussians
Variational training of mixture models and topic models with Multinomial likelihoods on simple toy “bars” dataset.
01: Standard variational training for mixture model
02: Training DP mixture model with birth and merge proposals
03: Standard variational training for topic model
04: Training HDP Topic Model with merge proposals
05: Training for HDP Topic Model with birth and merge proposals
01: Standard variational training for mixture model
Variational training of mixture models and topic models with Multinomial likelihoods on toy “bars” dataset with multiple bars per document.
01: Standard variational training for mixture model
Text from some restaurant reviews (Taddy 2012). Variational training of mixture models and topic models with Multinomial likelihoods.
VB coordinate descent for Mixture of Multinomials
VB coordinate descent for DP Mixture of Multinomials
Standard variational training for topic model
Birth and merge variational training for topic model
Text from a few thousand wikipedia articles. Variational training of mixture models and topic models with Multinomial likelihoods.
Scalable training of HDP topic models
Birth and merge variational training for topic model
Variational training of mixture models and HMM models with various Gaussian observation models.
Comparing models for sequential data
Visualizing learned state sequences and transition probabilities
Visualizing learned state sequences and transition probabilities
Topic models with Gaussian likelihoods